Posts Tagged ‘grid architecture’

The Hanukkah Armadillo and predictions for 2010.

18/12/2009

PL Junior is three years old and a bit, and I thought I’d share an interesting conversation from Friday last week, which was the first night of Hanukkah, whilst I was driving PL Junior to school.

‘Daddy, when will it be dark?’

‘Not until this evening, I’m afraid.  Why do you ask?’

‘Is tonight Hanukkah?’

‘Yes, it is …as soon as the sun goes down.’

‘When it gets dark we light the candles?’

‘Yes, are you excited?’

‘Um hmmm!  After the candles the Hanukkah Armadillo will come and sing me a lullaby.  And bring me presents!’

I start to laugh.  PL Junior isn’t.  He’s deadly serious.

‘Erm, sure. I suppose.‘  Not good to hurt a three year old’s feelings.  Best play along.

‘What else does the Hanukkah Armadillo do?’

‘He has flying powers, just like rocket!  And big jet!  And he comes down with presents each night of Hanukkah!  But he doesn’t like biscuits.  Only water and cucumbers.’

I dropped off PL Junior at school and then phoned Mrs. PL where, after we got done laughing, she tells me that she was channel hopping recently when she happened upon the Friends episode with the Holiday Armadillo.  It being one of her [and my] favourites, she watched and laughed.

PL Junior didn’t.  My father lives in Texas.  PL Junior thought it was real.  So, being a great mum, Mrs. PL decided to just go with it.  And thus was born the legend of the Hanukkah Armadillo in Casa PL.

My father just sent me an email offering to send a stuffed armadillo to place next to our menorah.

Thanks dad.

What’s this got to do with Data Storage & Protection?

One of the many things that I love about living in the UK generally and London specifically is the amazing diversity and breadth of opinion.

I don’t know if there is such as thing as a Universal Truth other than to say that everyone I know and work with would describe family as being paramount.  It doesn’t really matter much if you celebrate Christmas, Hanukkah, Diwali, Eid, Kwanzaa, Christmaskah, or Festivus …or none of these …you tend to do what you do because it makes you and your family happy.  And, frankly, how you observe or celebrate will often vary from family to family …who’s to say if there’s a ‘right’ way or ‘wrong’ way to do it?

Although I must say, there is a lot of merit in the Airing of Grievances around the Festivus pole.  But I digress.

As I’ve said before, I have a religion and it ain’t storage.  I’m not aware of any ‘one size fits all’ solution, but we have developed an equation and methodology which gives us as fair and equitable a way as I know for selecting and recommending solutions.  Just as families celebrate holidays in different ways so too do customers select and implement technologies.

That said, my predictions for 2010:

1.  The emergence of the virtualised datacentre.

What do you get if you add optimised storage, deduplication, storage compression, converged networks, virtualised servers, virtualised applications, integrated backup to disk/replication/tape storage, automation for the whole lot …I’m probably missing a few bits, but you get my point.  Whether you call it VCE Vblock, IBM Dynamic Infrastructure, NetApp Virtualise Everything, HP Converged Infrastructure, they’re all articulations of a virtualised datacentre.  We’re headed towards integrated and virtualised datacentres and one of the fascinating things about IT is stick around long enough and you’ll see the same ideas come back round.  We deconstructed the mainframe in pursuit of open systems, but now open systems are creaking under the load of data proliferation and low utilisation.  What to do?  Well, reconstruct the mainframe conceptually using integrated open systems.  And that’s what customers want …not necessarily flashy marketing schtick, but, rather a self healing/self tuning, policy based datacentre which allows customers to flex up/down depending upon market conditions with predictable and best costs and then sits invisibly in the corner and behaves!

2. Innovative products continue the march to becomes universal features.

There was a time when thin provisioning, data deduplication, disk spindown for underutilised disks, and automated storage tiering were products in their own right.  But if a product is so great that it demonstrably reduces costs and increases utilisation, shouldn’t it be a feature?  Exactly.  Which is why 2009 saw EMC, IBM, NetApp, HP, and HDS introduce or extend these technologies as features in their product sets.  And I don’t think that the list will stop there as I expect to see data compression enter as a feature in primary storage in 2010 as well.  Why should we care?  As features within an infrastructure we can use these technologies holistically, whether in a ‘branded’ virtualised datacentre or one composed of open products which work together seamlessly, to deliver storage at the most appropriate cost from creation to cremation with zero disruption to production business.

3. Server and desktop virtualisation are no longer incorrectly viewed as workloads.

VMware and Microsoft Hyper-V are good examples of server virtualisation, whilst VMware View and Citrix XenDesktop are good examples of desktop virtualisation.  Virtualisation of physical resources undoubtedly delivers higher utilisation and cost benefit, however virtualisation is a technology not a workload.  What’s the difference?  A workload is what you do with the technology, and everyone will do something slightly different with their workload based upon their business needs.  If you forget that and leave out the storage design and tuning, you may fail to achieve cost benefit at best or have to abandon the project at worst.  Don’t believe me though, have a read of this blog post by Chad Sakac, worldwide VP of VMware and all round virtualisation guru who is intimately involved with such deployments worldwide.

4. Grid storage / scale out storage take hold and never look back.

IBM XiV, EMC Vmax, NetApp ONTAP v8 …don’t look now, but everybody’s going grid.  Will it solve global warming, introduce world peace, and cuddle baby seals?  Not exactly.  But it will introduce self healing/tuning highly efficient and utilised storage at attractive price points using commoditised components …and become the bedrock of the virtualised datacentre.

5. Automation, automation, automation.

What use are virtualised servers, virtualised desktops, virtualised applications, self healing/tuning storage systems if you have to provision and manage them manually?  Precisely.  If you can’t automate it, chances are it will be left behind in the march towards the virtualised datacentre.  But don’t stop there as pure automation should give rise to …

6. Automation gives birth to policy based storage.

Remember aligning data to business value?  Of course you did!  Automation will remove the manual nature of managing IT and related resources, but customers don’t just want a conveyer belt of chocolates with Lucy and Ethel.  Automation which allows us to align a customer SLA and/or business value is what we truly want to deliver …policy based movement of data with zero disruption.

7. Customers continue to interact openly and publicly.  Are we listening?

Customers are speaking with us and our vendor partners publicly and openly in ways that we never could have anticipated through the use of technologies such as Twitter and blogs.  If you think that Twitter is just a platform for people to tell you what they’ve had for breakfast, you’ve missed the point entirely.  These platforms are a way for end users/customers to interact with one another as well as business partners efficiently. Storagebod, an end user at a major media company, took the time to write six individual letters to Father Christmas regarding what he would like to see from vendorsIanhf, an influential end user at a major telecoms company, took the time to write about what he expects from business partners.  Think you’re Elite?  See if you can get through all 17 of Ianhf’s points saying ‘yes, we/I do that’.  I’m not sure that I can, and that’s the point …by actively reading and engaging with what our customers are saying we and vendors have two choices …ignore it and continue making products / delivering services which customers find shoddy, or listen and strive to evolve and provide the products / services / solutions which make a difference to them in a positive way.  I know which one I’ll be choosing in 2010, otherwise we’re left just talking to ourselves and what fun or use is that?  Ignore customer public sentiment at your peril, as it would seem AT&T did in the USA with iPhone users.

8. England win the World Cup, Watford FC are promoted to the Premiership, and Mrs. PL stops yelling at me for snoring.

Okay, I kinda made that last one up but a PL can dream can’t he?  And who knows what 2010 will bring but what I do know is this …England, Watford FC, indeed all of us are in control of our destiny for the year ahead.

Have a great Christmas / Hanukkah / Diwali / Eid / Kwanzaa / Christmaskah / Festivus …none of the above …enjoy your holiday with your family and see you in 2010.

I’m off to light the last Hanukkah candle with Mrs. PL, PL Junior …and anxiously await the arrival of the Hanukkah Armadillo.

-Matthew

Click here to contact me.

Advertisements

On vendor agnosticity …and being selective.

06/09/2009

When I was growing up, I had an Uncle Malcolm.  Now …nothing unusual in having an Uncle Malcolm, save Malcolm isn’t a very common name in the United States …and he wasn’t my uncle.  To be sure, Malcolm was anything but common …some would call him an eccentric here in the UK, whereas most people in the States called him ‘weird’ …and my father insisted I call him Uncle Malcolm out of respect.  You see, Malcolm and my father worked together …and when I was a kid I thought Malcolm was the coolest guy in the world.  A mainframe programmer who has remained single his entire life, Uncle Malcolm taught me much of what I know of small plane flight and also how to play ‘Star Trek’ on the mainframe.  I spent a lot of time with Uncle Malcolm when my father brought me in to the office with him on weekends to verify mainframe backups and the like, and Malcolm was a bit of a minor rock star in my father’s company …there wasn’t much Malcolm didn’t know about mainframes, and what he didn’t know probably wasn’t worth knowing.

And then a young upstart named Bill joined my father’s department …indeed, he has been recruited and hired by my father as dad was beginning to develop solutions to his corporate ‘open systems’ requirements.  You see, the young upstart was …a client / server engineer!  GASP, egads!  It is de rigeur and not uncommon now, but back then?  Well, let’s just say that Bill and Malcolm didn’t exactly see eye to eye!  Bill didn’t do himself any favours by calling mainframes ‘dinosaurs‘ …but Malcolm didn’t exactly take the high road either when he continually referred to Bill’s servers as ‘not REAL computers’.  Things somewhat deteriorated after that, and really came to a head when Uncle Malcolm caught me playing Microsoft Flight Simulator on an IBM XT PC that Bill had setup for me.  In the end, my father made it quite clear to both of them that he wasn’t about to buy flak jackets and UN blue helmets …get on or get out …and thus began a computing cold war which carries on between them to this day.

What has this got to do with data storage and protection?

The problem with Bill and Malcolm was that they were both right …and both incredibly wrong.  Bill was right to highlight that the open systems movement was the evolution of corporate computing …but mainframes were hardly dinosaurs waiting for a hurtling comet to wipe them out forever.  Malcolm was right to highlight the incredible uptime and reliability of mainframes …but open systems were ‘real’ computers and offered corporate users options and flexibility that mainframes simply don’t.  And frankly …who cares?  Their job was to understand how to solve business issues without bringing computing religion into it …and they both failed …miserably.  As a strange circumstance, I got to work with both Bill and Malcolm after my father moved to Texas to take a CIO position and his previous company needed help completing a three tier client / server implementation …in plain English, an open systems infrastructure for their order and delivery system which leveraged the mainframe on the back end and brought the best of both worlds to their corporate users.  Bill and Malcolm called a truce, and I bump into Bill now and again at industry conferences …and Malcolm is making money hand over fist as one of the few people who still know how to reliably make billing systems work on mainframes.

Over the past twelve or so months, some vendors have described me at times …quite unhelpfully and inaccurately, as it happens …that I am either ‘in love’ with a competing vendor product at best, or a [insert competing vendor solution] ‘bigot’ at worst.  Now …they do have one thing right in that equation.  I am a bigot.  I am a customer bigot and …as I’ve stated before …I have a religion, and I can assure you that it is not storage.  Every vendor I talk to is rightly proud of their solutions …and their job is understandably to tell the world that their storage is the only solution to solve customer problems …but the simple fact is that each vendor solution is applicable and ‘correct’ depending upon the customer requirements.  And I’ve not met two customers yet who both had the exact same requirements.

I’m a customer bigot and am only truly interested in how data storage and protection can help our customers both save money as well as remain competitive in their respective markets.  Okay, now and again I will feed my inner geek and get into esoteric conversations with either our consultants or vendor partners by discussing the merits of NAS and NFS/CIFS with traditional database and OLTP systems …or how stable grid storage systems which retain thin provisioning and zero page reclamation in the frame with universal fibre channel, iSCSI, CIFS/NFS likely represent the future of data storage.  But this is not a message likely to excite a customer …customers spend money to solve problems, and nothing of what I’ve written in the last two sentences gives any hint as to how these systems would do this.

My job …and, by extension, the job of our consultants …is to use our consultancy equation* to evaluate our customer’s needs and requirements with a view to recommending what they need, not necessarily what our vendors have.

Please don’t misunderstand my messaging here …this isn’t a vendor bash, and we rely on our vendors to continue to make great products.  But just like Bill and Malcolm were convinced that they were both right, the simple truth was that there were corporate needs which open systems solve and corporate needs which the mainframe solve.  How much better and robust their solution could have been had they decided to work together as opposed to slagging one another off.

Our solutions are designed to solve customer issues as opposed to highlighting the ‘speeds and feeds‘ of a one storage solution over another.  Will thin provisioning solve the problem?  Perhaps, but should we access the storage via fibre channel, iSCSI, or CIFS/NFS?  And what about data deduplication …will that help?  Probably, but should we be looking at inline or post processing data deduplication …or both?

The only way to know is to listen to our customers and articulate our solutions in a way that makes it very clear how our solution helps to solve that issue.  Please don’t hesitate to  contact me if I can be of any assistance in helping you take our customers on this very important journey.

Have a great weekend.

-Matthew
Click here to contact me

*The consultancy equation we use is [ROI] + [CBA] + [Disruption to Production Business] = Composite Solution Score …return on investment plus cost benefit analysis plus potential disruption to the customer’s production business to give us a composite solution score where we can fairly and accurately measure multiple vendor solutions.  It’s somewhat complicated, but feel free to contact me and I will happily walk you through how this works in practice.

1984 revisited.

28/08/2009

I remember watching the Super Bowl XVIII with my father in 1984 when, unexpectedly during an otherwise boring third quarter, an advert came on which would change the way we think about computing forever.  The advert was directed by Ridley Scott …who would go on to make such classics as Blade Runner, Alien, and Gladiator …cost millions of dollars and months to make, lasted exactly sixty seconds, and was shown exactly …once.  I remember watching the advert speechless and wishing to see it again …but I wouldn’t see it again until many years later when You Tube made such things possible.

I won’t go into the advert in any great detail as it deserves to be seen and digested [click here to view], but it purported to introduce the Apple Macintosh to the world on 24 January 1984.  What it actually did was fire the first salvos of the ‘open system’ movement against the traditional mainframe world.  My father, who was an executive in the ‘data processing’ department …Information Technology, or ‘IT’ didn’t exist yet …watched the advert, turned to me, and proceeded to tell me how we would someday watch movies without going to the cinema, watch any movie ever made when and where we wanted to, listen to any piece of music ever recorded when and where we wanted to …and carry the sum total of the world’s library content in our pockets.

Now, my old man was heavy into Star Trek [still is, as am I, and I’ve passed on this geekdom to PL Junior!] and I have to admit that, at the time, I thought he had been watching one too many episodes with JT Kirk and the boys and/or had had one too many G&Ts during the course of the Super Bowl.  I nodded politely and went back to noshing on the nachos and sour cream my mum had made.

What has this got to do with Data Storage and Protection?

Far from being doolally, my father was describing things to me in 1984 that we now take for granted.  iPods, You Tube, Spotify, e-ink eReaders …all have become reality and, it is projected that within 15 years we will have personal storage systems capable of holding the contents of all of the world’s libraries in a form factor small enough to fit in your pocket.  Really.  But how did my father predict such things?  Was he a futurist who missed out on the big time?  Sadly, no.  He went to uni with people who would go on to work for DARPA and they would often sit up late into the evening over G&Ts discussing the latest developments in data networking technology.  They could see the practical applications of the internet they were developing, and I would argue that there are many parallels to be found in the success of the internet and the future of data storage.

The internet was not originally designed to be the delivery mechanism for fine purveyors of pornography but, rather, a way for the US government and military to communicate from coast to coast and all throughout the US in the event of WWIII …and the Soviet Union had wiped entire communication nodes …and cities …off the map with nuclear weapons.  It is a system that is designed to fail …a resilient system which can continue to operate even after multiple and massive failures.  Another interesting feature of the internet is it is designed to use standard ‘off the shelf’ components such that, as the components’ quality and processing power increases, they can be put into the internet without having to take down or redesign the whole flippin’ thing.  Think dial up modem to ‘wired’ broadband to ubiquitous WiFi and you get the picture.

A similar revolution is happening in storage.  The original storage arrays were really more like massive servers on steroids …in fact, they looked and behaved much like the mainframes they were meant to replace.  They have a central processor known as a controller, cache which acts much like the memory in a server, and well …disks.  The disks provide the massive and shared storage, but are connected to the controllers and cache.  A great architecture to start with, but as the amount of data we are creating has exploded exponentially it has become less and less efficient and more difficult to manage.  Indeed, whilst functionality has been added to the architecture …data replication, modular arrays, iSCSI, NAS, and so on …the central principle of architecture design hasn’t changed all that much in quite a while.  Most importantly, they are designed to never fail …and as a result of that principle, we find ourselves in something programmers call an ‘infinite loop’ which becomes ever more expensive to manage.

So what’s the answer?  Well, we need to turn the central principle on it’s head and design systems which fail.  Now, I know that may seem counterintuitive, but bear with me.  The quality of standard components has increased significantly over the past ten years …we now have hard drives and Intel processors in our laptops which would have powered supercomputers ten years ago …but, things still fail from time to time.

My view is parallel processing or ‘grid architecture’ storage is the answer, and these systems will soon eclipse the traditional storage architectures.  What is grid architecture storage?  The specifics can be somewhat complicated, but in principle you have data and communication modules which replace the controllers and cache with software to connect all of the modules together.  What does this give us?  The ability to use algorithms which write data to all of the data drives simultaneously …and as each module uses standard components with CPU and memory on board, I can lose a module without losing data as well as increasing system performance by replacing the standard components with the new whizz bang models as they become available.  The secret sauce of the software which connects the modules is what allows us to provide performance and reliability equivalent or greater to traditional systems …at an acquisition and management cost much lower than traditional storage systems.

So who is using grid architecture?  If you have Googled anything recently, you’ve used grid storage architecture.  And, if grid storage is the answer, who makes it?  Well, IBM XiV was the first past the mark but now EMC Vmax has joined the scene …and NetApp, HDS, HP aren’t far behind.  What will set each of them apart, in my opinion, is how they implement grid architecture as well as the functionality they provide as standard.

If you aren’t thinking about grid architecture storage, you should be …this is a tidal wave which is already transforming the storage marketplace and, whilst we  have ‘first mover’ advantage with the promise of fair pricing at the moment this won’t last forever.

Please feel free to contact me if I can help you understand grid architeture more fully and in greater depth, including the very real cost savings which can be gained.

-Matthew
Click here to contact me